Open-source Marketing Mix Modeling (MMM) received a new addition last week with Google’s release of Meridian. Meridian will replace Google's unofficial lightweight MMM solution in the coming months. - Google's Meridian: https://lnkd.in/esJS-yWx - Lightweight MMM: https://lnkd.in/erEhvQW5 Alongside Meridian, the other two major open-source MMM initiatives are: - Meta's Robyn: https://lnkd.in/e2ApWyVd - Uber's Orbit: https://lnkd.in/eRiEtgQ3 Open-source MMMs are making it easier for brands and agencies to dive into building their own MMMs. But keep in mind: ✋ MMM is a complex media measurement solution – just like owning a hammer and chisel doesn't make you a skilled carpenter, simply using these libraries doesn’t guarantee robust/ successful media measurement 🧠 I recommend initially working with an experienced consultancy or agency to implement your MMM projects. At MCSP, we offer this support for our clients. 🔥 Once your MMM framework has demonstrated its impact on media budgeting and optimization, you can then think about in-housing – but there is no rush if things are going well. Media measurement is evolving, and MMM will be a key player. However, if done badly, it can cause your key stakeholders to lose faith in the approach and potentially lead to poor media investment decisions. 🏇 Get started with MMM as soon as possible, but look for experience rather than quick fixes.
Perfect advice. I also penned a post a few days ago on the hidden costs of open source MMM tools.
Another example of MMM's high cost being disrupted. Great for marketers if it's deployed correctly and with skepticism - loved what you said about these models not guaranteeing successful measurement on their own.
I wonder if Google will open source Meridian’s model, would be cool to see how they handle causality
Another great open source alternative for MMMs is PYMC-Marketing. Generally, one the best I've tried
MMM is best when tailored to unique business needs and data availability! Open libraries are a great way to start but cannot guarantee success.
Senior Advisor - Strategy, PE VCP - Pricing, Go-To-Market, Analytics
4moMMM is hot. Disruptive open source “AI” tools like Robyn and Meridian are exciting. There’s some defensive language from incumbents (there always was). Of course do modelling properly, of course work with experienced MMM analysts. But people shoukd be trying out new things, give it a whirl. This is potentially going to herald an amazing revolution, new gates on the baptistry. Let’s embrace the new and what we can potentially learn. Some MMM econometricians are arguing against these AI tools. But they would, wouldn’t they? I was part of the Edge team that created a first generation of automated modelling tools 20 years that begat the MMM software that followed at Accenture, SQED, MASS etc. And that tech influenced the workflow tools used by all the major MMM houses since. At that time naysayers said blah blah blah it’s all about the art of the econometrician, your automated tool won’t work. Well it did work! We sold the company to Accenture. The Accenture diaspora went on to software houses and agencies who build on those ideas. And we created a new MMM software industry, we built new agency tools etc. That experience taught me that people can be resistant to change initially, but new tech can be a catalyst to better ways.